Jerry Zhang <jerry@xxxxxxxxxx> writes: > Unrelatedly would you have context on why --3way falls back on 3way > rather than trying 3way first then falling back on apply_fragments if > blobs don't exist? Historical accident, following the order in which these features were invented, plus applying a single patch straight has been faster than doing a 3-way. I tend to agree with what you are hinting at, though, as I do not offhand think of a situation in which a successful 3way merge would be less correct than a straight patch application, and if the user explicitly has told us to do "--3way", that is a sign that it is acceptable to try 3way first even if it costs more cycles. Those who has been giving "--3way" from inertia would notice if the performance difference is large enough and may complain, though.
